【作者】 郝劲波；

【导师】 忽满利；

【作者基本信息】 西北大学 ， 光学， 2007， 硕士

【摘要】 光学系统对二维图像进行识别具有非接触、速度快、并行处理和平移不变性等优点，从而得到了广泛应用。但光学系统只能对二维信息进行处理，不能直接实现三维物体的识别，而必须首先将三维物体的信息转换成二维信息。本文提出一种基于微透镜阵列多视角成像特点，将三维物体的深度信息转化为二维的透射像阵列的角度信息，利用光学二维图像识别技术，实现对三维物体识别的方法。并对识别过程进行了理论分析和计算，用匹配滤波的方法实现了对三维物体骰子的实时识别。实验结果表明，本方法的相关识别能力较高，并且具有很强的灵活性，对于有微小旋转、微小平移的三维物体也可进行识别。在该方法的基础上，接着讨论了其对于具有小角度旋转的三维物体的旋转不变识别，并进行了实验验证，得到了比较好的识别效果。并且实现了实时识别，使该方法的应用范围进一步扩大。随后对得到的透射像阵列图像，应用综合判别函数，用计算机进行模拟实验，对三维物体骰子的相同面、不同面旋转实现了旋转不变识别，得到了比较好的识别效果，从而得到一种新的对三维物体进行旋转不变识别的方法。更多还原

【Abstract】 Because of non-contact, speed, parallel processing and shift invariant, two-dimensional(2-D) object recognition by use of optical system is be used abroad. But optical system only can be used for processing of 2-D information, not be used for 3-D object recognition, so 3-D object information must be transformed for 2-D information firstly.A novel method for three-dimensional(3-D) object recognition by use of microlens array is proposed. The conversion of depth information of 3-D objects to 2-D elemental images by use of the characteristic of microlens array makes it is possible to recognition by optical processing. Correlative analysis and calculations has been done for the process of recognition, and real-time recognition of dice ,a 3-D object ,has been accomplished by use of the method of match filtering. The result showed that the proposed method performed better, and it can be used for recognition of slightly rotated and slightly shifted object flexibly.3-D object micro-angle rotation invariant recognition is discussed continually, The experiment result of 3-D object real-time recognition showed that the proposed method performed better, and then, A novel method for 3-D object rotation invariant recognition by use of synthetic discriminant function(SDF) is received. The computer simulate experiment for rotation invariant recognition for dice, a 3-D object has been accomplished.更多还原